A century and a quarter of Canadian parliamentary debate

Antisemitism in the Canadian Hansard

Searched from 1901 to 2026. Every statement in the House of Commons, the Senate and parliamentary committees containing antisemitism and related terms, in both official languages, with the passage and a link to the record. The word does not appear in the House until 1934: the first thirty-three years return nothing at all.

A project of the Lab for the Global Study of Antisemitism, Anne Tanenbaum Centre for Jewish Studies, University of Toronto. Ron Levi. · Built, analyzed, and annotated with Claude Code.

Loading…

$ data & code — R, Python, curl, JSON, citation

static JSON · CORS open · no key · no rate limit

# the chart follows your query, so a search is also a frequency-over-time plot
library(jsonlite); library(dplyr)
base <- "https://hansard-antisemitism.pages.dev/data/"

hansard <- bind_rows(
  mutate(fromJSON(paste0(base,"instances_lipad.json"),     flatten=TRUE), chamber="House"),
  mutate(fromJSON(paste0(base,"instances_house.json"),     flatten=TRUE), chamber="House"),
  mutate(fromJSON(paste0(base,"instances_senate.json"),    flatten=TRUE), chamber="Senate"),
  mutate(fromJSON(paste0(base,"instances_committee.json"), flatten=TRUE), chamber="Committee"))

# what each party does with the term
hansard |> filter(!procedural, !is.na(party)) |>
  count(party, act) |> group_by(party) |> mutate(share = n/sum(n))

# sitting days per year: the unit that survives the 1994 change of source
hansard |> filter(!procedural) |> distinct(chamber, date, year) |> count(chamber, year)

A real R session, running in your browser through webR. Nothing is sent anywhere. The corpus is already loaded as hansard, a data frame of — rows. First run downloads R itself, about 30 MB.

R not started

Press Run. The first run takes a moment while R downloads.

# pandas
import pandas as pd
base = "https://hansard-antisemitism.pages.dev/data/"
panes = ["house", "lipad", "senate", "committee"]
df = pd.concat({p: pd.read_json(base + f"instances_{p}.json") for p in panes},
               names=["pane"]).reset_index(0)
df = df[~df.procedural]

# orders of worth by party (Boltanski & Thevenot)
df.explode("orders").groupby(["party", "orders"]).size().unstack().pipe(
    lambda t: t.div(t.sum(1), axis=0)).round(2)

# everything, as one stream
for p in house lipad senate committee; do
  curl -s "https://hansard-antisemitism.pages.dev/data/instances_$p.json"
done | jq -s 'add | map(select(.procedural | not))' > hansard.json

# count speech acts
jq 'group_by(.act) | map({act: .[0].act, n: length}) | sort_by(-.n)' hansard.json

@misc{levi2026hansard,
  author = {Levi, Ron},
  title  = {Antisemitism in the Canadian Hansard},
  year   = {2026},
  note   = {Lab for the Global Study of Antisemitism, Anne Tanenbaum Centre for Jewish Studies, University of Toronto},
  url    = {https://hansard-antisemitism.pages.dev}
}

Parliamentary text is reproduced under the Speaker's Permission; cite the Debates of the House of Commons or of the Senate as the source of any passage.

8.4Mspeeches and statements searched

→

606Mwords in the 1901–2019 House record alone

→

2,837statements that name antisemitism

Roughly one statement in every 3,000 spoken in the Parliament of Canada. The search runs from 1901, but across the first thirty-three years and 95 million words the word never appears once. Every statement that does is here, with its passage and a link to the official record.

Work in progress. Corrections, suggestions and advice are very welcome: ron.levi@utoronto.ca

A statement appears here only if it uses the word antisemitism or a close variant. Speech that attacks Jews without naming them — through terms like “globalists”, “international bankers” or “cultural Marxism”, which carry the meaning while leaving it deniable — is not in this corpus at all. Why, and what it would take to capture it.

Coding Methods

New to Hansard? Start here

Hansard is the official report of what is said in Parliament. The name comes from Thomas Curson Hansard, a nineteenth-century London printer, and it is used across the Westminster parliamentary world: the United Kingdom, Canada, Australia, New Zealand and others each publish their own. It is close to verbatim, lightly edited for repetition and grammar but not for content, and it is published for every sitting day.

This site covers the Canadian federal Parliament in Ottawa.

House of Commons	The elected lower chamber, currently 343 seats. Members are called MPs, each elected from a local district known as a riding. This is the main tab.
The Senate	The appointed upper chamber, 105 seats. Senators are appointed rather than elected and serve until age 75. The Senate reviews and can amend legislation passed by the Commons. Included here from 1996, which is as far back as the Senate publishes online.
House committees	Committees of the House of Commons. Parliament does most of its detailed work in them: small standing groups of MPs, each covering a subject area such as justice, immigration or foreign affairs. They examine bills line by line, run studies, and call witnesses to testify. Their transcripts are published separately from the Chamber's and are called evidence. Two things follow. Much of what is said in them comes from invited witnesses, including community organisations, academics and police, rather than from politicians. And a committee can dwell on one subject for weeks, so a single study can produce far more statements than a day in the Chamber. They are kept in their own tab for both reasons. Senate committees are not included, since the source used here covers Commons committees only.

The main federal parties you will see: the Liberal Party and the Conservative Party (and its predecessors, including the Canadian Alliance and Progressive Conservatives), the New Democratic Party on the left, the Bloc Québécois, which runs candidates only in Quebec and favours Quebec sovereignty, and the Green Party. Historical parties appear in the earlier records, among them Social Credit and the Co-operative Commonwealth Federation, the NDP's predecessor.

English and French are both covered. Canada has two official languages, and members speak in either; Hansard publishes both versions of every intervention. Speech delivered in French is therefore included throughout, and is usually found through the English text that accompanies it. French-specific vocabulary is searched directly as well, so antisémitisme, antijuif, antijuive, judéophobie and haine des Juifs all match. Records found only in the French text carry a matched in French tag, and you can isolate them with the query matched:fr. Note that a French tag marks where the search term was found, not the language the member spoke, since most French speech is caught through its English rendering.

What this site is. Every statement containing antisemitism or a related term, found by automated search rather than by editorial judgment. A statement appears because it contains one of the words. It has not been assessed for whether it makes, denies or disputes a claim about antisemitism, and the corpus includes speech that is itself antisemitic.

the series

Statements per year

Click a bar to filter the list below. Click again to clear.

Search words Speaker Party Sort

Query syntax

`boycott divestment`	both words, anywhere in the record
`"hate speech"`	exact phrase
`-Israel`	exclude records mentioning Israel
`party:Liberal`	field search: speaker, party, riding, province, topic, term, text
`year:2015` · `year:2015-2020` · `year:>2020`	a year, a range, or an open bound
`term:judeophobia`	which matched term brought the record in
`where:heading`	matched the debate heading, not the speech
`matched:fr` · `matched:en`	which language text the term was found in
`source:lipad` · `source:openparliament`	which database the record came from
`party:Bloc OR party:NDP`	alternatives
`speaker:Cotler year:>2010 -Israel`	combine freely; terms are AND by default

coded findings

Patterns in the coded data

Percentages are shares within each period, so periods of different size stay comparable. Selecting a band in one chart narrows the other three, but never the chart you clicked, so you can still see where your selection sits.

What the speaker does with the term, by period

Grounds of the claim, by period

Named alongside antisemitism, by period

What should be done, by period

Coverage: the House searched from 1901, and the years still missing

The corpus is still being assembled. Hatched years have been requested and are not in the tool yet.

in the tool requested, not yet included outside the project

Compare parties, chambers and periods side by side

Compare on Rows are shares within each group, so groups of different size stay comparable.

Break these results down by party, speaker and sitting day

How this was built, and what it does not show

a project of

Lab for the Global Study of Antisemitism

This site is a project of the Lab for the Global Study of Antisemitism, housed in the Anne Tanenbaum Centre for Jewish Studies at the University of Toronto. The Lab supports research, teaching and events on the causes, manifestations and effects of antisemitism across the humanities, social sciences and professional fields.

Ron Levi. Distinguished Professor of Global Justice, Munk School of Global Affairs and Public Policy and Department of Sociology, University of Toronto.
ron.levi@utoronto.ca

Correspondence about this corpus, corrections, and requests to reuse the data are welcome.